Asthma afflicts millions of people worldwide. Genetic and environmental influences on objective intermediate asthma phenotypes in Dutch twins

Transcription

1 Eur Respir J 2010; 36: DOI: / CopyrightßERS 2010 Geneti and environmental influenes on objetive intermediate asthma phenotypes in Duth twins T. Wu*, H.M. Boezen #, D.S. Postma ", H. Los +, P.E. Postmus 1, H. Snieder* and D.I. Boomsma e ABSTRACT: It is unlear to what extent the same set of environmental or geneti fators regulate objetive intermediate asthma phenotypes. We examined heritabilities of these phenotypes and estimated their environmental and geneti overlap. We studied baseline lung funtion (fored expiratory volume in 1 s (FEV1), fored vital apaity (FVC) and FEV1/FVC), bronhial hyperresponsiveness, number of positive skin prik tests (SPT) to 11 allergens, serum total immunoglobulin (Ig)E, number of positive speifi IgE tests to four allergens and eosinophil ounts. 103 twin pairs were studied (46 monozygoti and 57 dizygoti; mean age: 22.5 yrs, range: yrs). Univariate and bivariate geneti analyses were performed after adjustment for signifiant ovariates. All intermediate asthma phenotypes showed signifiant heritabilities (47 83%). Most phenotypes were substantially orrelated, whih was mainly due to shared geneti fators. Pairs of phenotypes with the largest geneti orrelations were speifi IgE and SPT (0.98), and total IgE with speifi IgE (0.87), with SPT (0.72), and with eosinophils (0.62). SPT showed signifiant environmental orrelations with total IgE (0.65), speifi IgE (0.70) and bronhial hyperresponsiveness (0.44). Geneti effets explain the majority of the variation in objetive intermediate asthma phenotypes. Additionally, orrelations between pairs of these traits are also mainly explained by geneti rather than environmental fators. KEYWORDS: Asthma, environmental orrelation, geneti orrelation, multivariate, twin Asthma afflits millions of people worldwide and is aused by multiple geneti and environmental fators. There has been a strong interest in searhing for suseptibility genes for asthma, driven by the prospet of better disease prevention, diagnosis and treatment. However, studies on the genetis of asthma are ompliated, sine there are diffiulties in standardising the diagnosis of asthma [1, 2]. Most urrent geneti studies have onentrated on asthma assoiated with atopy, defined in a number of different ways: by elevated total serum immunoglobulin (Ig)E levels and/or positive skin prik tests (SPT) to one or more allergens. The binding of >IgE to its reeptors results in mast ell ativation and eosinophil reruitment, another asthmarelated phenotype. Bronhial hyperresponsiveness, the inreased bronhoonstritor response to nonallergi stimuli is a pre-requisite for an This artile has supplementary material available from asthma diagnosis [3]. These measurable and biologial markers, so-alled intermediate phenotypes, suh as IgE levels, SPT, eosinophils and bronhial hyperresponsiveness, underlie the pathophysiology and linial expression of asthma [4]. They are more objetive, aurate, and more informative to use in geneti analyses than linial definitions or dotor s diagnosis of asthma. So far, only how far these intermediate phenotypes are driven by similar or different geneti or environmental fators has been studied. Twin and family studies are widely used to estimate geneti and environmental ontributions to atopi disease [5, 6]. The multivariate lassial twin design an help to estimate the degree to whih the same geneti and environmental fators influene different intermediate phenotypes [7]. If there is overlap in genes for two traits AFFILIATIONS *Unit of Geneti Epidemiology and Bioinformatis, Dept of Epidemiology, University Medial Center Groningen, University of Groningen, # Unit Chroni Airway Diseases, Dept of Epidemiology, University Medial Center Groningen, University of Groningen, " Dept of Pulmonology, University Medial Center Groningen, University of Groningen, Groningen, + Dept of Pulmonology, Nij Smellinge Hospital, Drahten, 1 Dept of Pulmonary Diseases, VU University Medial Center, and e Dept of Biologial Psyhology, Vrije Universiteit, Amsterdam, The Netherlands. CORRESPONDENCE H.M. Boezen Unit Chroni Airway Diseases Dept of Epidemiology University Medial Center Groningen University of Groningen Hanzeplein 1 PO Box RB Groningen The Netherlands h.m.boezen@epi.umg.nl Reeived: Aug Aepted after revision: De First published online: Jan European Respiratory Journal Print ISSN Online ISSN EUROPEAN RESPIRATORY JOURNAL VOLUME 36 NUMBER 2 261

2 ASTHMA T. WU ET AL. it is expeted that the ross-twin ross-trait orrelation will be higher in monozygoti (MZ) twins than in dizygoti (DZ) twins for these traits. Using this information annot only widen our understanding of atopi omorbidity, it an also enhane gene-mapping efforts [6] and onsequently enable us to understand the interplay between different pathways underlying asthma, benefitting the disovery of targeted treatments and interventional measures. The aims of the urrent study were to estimate the relative influene of geneti and environmental fators on objetive intermediate asthma phenotypes and, more importantly, to what extent orrelations between these intermediate phenotypes an be explained by geneti and/or environmental fators using a sample of 103 young Duth twin pairs. METHODS Subjets The families partiipating in the asthma study were a sample of a larger number of Duth twin families who partiipated in the Netherlands Twin Register (NTR), an ongoing survey study of health-related behaviour [8 10]. The NTR represents a random sample of Duth families with twins, reruited from ity ounil registrations. We seleted families in whih twins were aged o18 yrs and who reported at least one member with a history of asthma. In total, 102 families partiipated, ontributing 103 twin pairs (one family ontributed two twin pairs) of whih 46 pairs were MZ and 57 pairs were DZ, of whih 26 pairs were of opposite sex (for details see the supplementary material). Zygosity was determined by DNA fingerprinting (Sequana, San Diego, CA, USA and TNO, Leiden, The Netherlands). The study was approved by the institutional review board and all subjets provided written informed onsent. Measurements We used the European Community Respiratory Health Survey questionnaire to define the symptoms presented in table 1 (e.g. asthma, ough, wheeze and dyspnoea, et.). Lung funtion test Lung funtion was measured using pneumotahographs (Vmax series, Sensor Medis Co., Yorba Linda, CA, USA) aording to the guidelines of the Amerian Thorai Soiety [11]. Eah twin pair was tested on the same day and at the same time of day. Fored vital apaity (FVC) measurement was followed by measuring of the fored expiratory volume in 1 s (FEV1). Three measurements were made until at least two satisfatory measurements were produed; the highest values being taken as baseline value. The ratio of FEV1 and FVC was used to determine the degree of airway obstrution. Bronhial hyperresponsiveness Bronhial hyperresponsiveness (BHR) was measured on a DeVilbiss 646 nebuliser (Devilbiss Co., Somerset, PA, USA) using the standardised protool desribed by COCKCROFT et al. [12], based on 2-min inhalations of doubling onentrations of methaholine of mg?ml-1 (see the supplementary material). The methaholine provoation onentration produing a 20% fall in FEV1 (PC20) was determined using linear interpolation of the last two points of the onentration-response urve [13]. Skin prik test Skin prik tests were performed using allergen-oated lanets for 11 ommon allergens: house dust mite, storage mite, mixed tree pollens, mixed grass, weed, at, dog, horse, hair, feathers and moulds. A partiipant was defined as atopi if at least one of these allergens tested eliited a mean wheal diameter o3 mm [14] and the negative ontrol had a mean wheal diameter,1 mm. Measurements of total serum IgE, speifi IgE and eosinophil ounts Measurements methods of serum total IgE and speifi IgE are desribed in the supplementary material. Peripheral blood eosinophil ounts were performed by flow ytometry. Analytial approah The aims of our analyses were two-fold. First, we estimated the relative influene of geneti and environmental fators on observed phenotypi variane in eight objetive intermediate asthma phenotypes separately and investigated sex differenes in geneti arhiteture. Seondly, we assessed to what extent phenotypi orrelations between these phenotypes an be explained by geneti and/or environmental fators. We used bivariate modelling to estimate ross-trait orrelations and to partition these into geneti and environmental omponents. Twin orrelations and heritability estimates Most phenotypes were adjusted for ovariates prior to the estimation of twin orrelations and model fitting. FEV1, FVC and FEV1/FVC were adjusted for age, sex, height and smoking. PC20 was adjusted for age and sex. Total IgE and eosinophil ounts were log-transformed to obtain normal distribution and both were adjusted for age and sex. Pearson orrelations were alulated for ontinuous traits. Polyhori orrelations were alulated for ordinal traits (i.e. PC20, SPT and speifi IgE) under a liability-threshold model [7, 15] (for details see the supplementary material). Although PC20 was measured as a ontinuous variable, it ould not be normalised in our study. Therefore, we reoded PC20 into an ordinal variable with six ategories. Lower reoded values represent higher atual values (e.g. 0 represented 160 mg?ml -1 ), indiating a lower bronhial responsiveness. SPT and speifi IgE were also defined as ordinal variables with six and four ategories, respetively, based on the total number of positive responses. Heritabilities for all traits were estimated using univariate geneti model fitting tehniques. Geneti modelling of twin data We used model fitting analyses of twin data [16], a tehnique based on omparison of ovarianes (or orrelations) in MZ and DZ twin pairs, whih allowed a separation of the observed phenotypi variane into its geneti and environmental omponents (fig. 1): additive or dominant geneti omponents and ommon or unique environmental omponents. The unique environmental omponent also ontains measurement error. Dividing eah of these omponents by the total variane yields the different standardised omponents of variane, for example, the heritability (h 2 ) an be defined as the proportion of the total variane attributable to additive geneti variation. 262 VOLUME 36 NUMBER 2 EUROPEAN RESPIRATORY JOURNAL

3 T. WU ET AL. ASTHMA TABLE 1 General harateristis and objetive intermediate asthma phenotypes of male and female twins Charateristis Males Females Sex effets p-value Subjets n Age yrs NS Height m ,0.001 Weight kg ,0.001 BMI kg?m NS Current smoking 26 (28.9) 27 (23.3) NS Asthma 47 (52.2) 61 (52.6) NS Cough 17 (18.9) 36 (31.0) NS Wheeze 24 (26.7) 38 (32.8) NS Dyspnoea 28 (31.1) 46 (39.7) NS Dyspnoea at night 5 (5.6) 24 (20.7) Mediation b-agonists 7 (7.8) 20 (17.2),0.05 Inhaled ortiosteroid 2 (2.2) 11 (9.5),0.05 Antiholinergi 1 (1.1) 1 (0.9) NS Objetive intermediate asthma phenotypes # FEV1 L ,0.001 FEV1 %pred NS FVC L ,0.001 FVC % pred NS FEV1/FVC ,0.05 Total serum IgE IU?mL ( ) 53 (21 172) NS Eosinophil ount ells?l ( ) 0.11 ( ) NS Methaholine PC20 mg?ml ( ) 41.8 (4 160),0.05 Positive SPT 30 (33.3) 49 (42.2) NS Positive serum speifi IgE 45 (51.1) 51 (47.2) NS Data are presented as mean SD, n (%) or median (interquartile range), unless otherwise stated. BMI: body mass index; FEV1: fored expiratory volume in 1 s; % pred: % predited, orreted for age and height; FVC: fored vital apaity; Ig: immunoglobulin; PC20: provoative onentration ausing a 20% fall in FEV1; SPT: skin prik test; NS: not signifiant. # : missing value for these phenotypes varied from 0 (0%, FVC and SPT) to 13 (6.31%, PC20). The existene of sex differenes in the influenes of geneti and environmental fators on the phenotype an take several forms [16] (see the supplementary material). A path diagram of the applied twin model is shown in figure 1; k is the salar fator that indiates that the total variane of the phenotype might differ between males and females. Bivariate geneti models In order to estimate geneti and environmental influenes on the assoiation between these intermediate asthma phenotypes, a bivariate Cholesky deomposition [16] was used to model the ovariane between any two of them (fig. 2). This model allows determination of the extent to whih the ovariane (or phenotypi orrelation (rp)) an be explained by geneti and/or environmental fators influening both phenotypes under study (see the supplementary material). The geneti orrelation (rg) between two traits gives an indiation of the amount of overlap between (sets of) genes influening those traits and ranges from -1 to 1. If rg is 1, then the same genes influene both traits. A orrelation of 0 indiates that both traits are affeted by a different set of genes. The positive or negative geneti orrelation reflets positive or negative assoiation between any two traits. Common and unique environmental orrelations between two traits are alulated in a similar way (fig. 2). In order to inlude both ontinuous and ordinal (PC20, SPT and speifi IgE) variables into the same bivariate model, we reoded ontinuous variables to ordinal ones with 10 ategories. Model fitting proedure Models were fitted to the raw data using normal theory maximum likelihood allowing inlusion of inomplete data (i.e. when data were only available in one twin of a pair). Depending on the orrelation pattern for MZ and the DZ twins, the geneti model fitting started with either an ACE (additive geneti, ommon environmental, unique environmental) model (if twie the DZ orrelation was larger than the MZ orrelation) or an ADE (additive geneti, dominane geneti, unique environmental) model (if the DZ orrelation was less than half the MZ orrelation). The signifiane of variane omponents, additive geneti, ommon environmental or dominane geneti, was assessed by testing the deterioration in model fit after eah omponent was dropped from the full model. Standard hierarhi Chi-squared tests were used to selet the best fitting models [16] in ombination with Akaike s information riterion (AIC 5 x2-2 degrees of freedom). The model with the lowest AIC reflets the best balane of goodness-of-fit and parsimony. EUROPEAN RESPIRATORY JOURNAL VOLUME 36 NUMBER 2 263

4 ASTHMA T. WU ET AL. r g r r g A C E A C E G G h e h e h 2 h 2 P1 P2 Trait 1 Trait 2 k 1 e 2 e 2 P1 P2 E E FIGURE 1. Path diagram for a univariate model. An opposite-sex twin pair is shown, twin 1 (left) being male and twin 2 (right) female. Observed phenotypes (P) for twin 1 and twin 2 are shown in squares, latent (i.e. unmeasured) fators are shown in irles: A: additive geneti fator; C: ommon environmental fator; E: unique environmental fator; r g is the orrelation between A, whih is 1 in monozygoti (MZ) twins and 0.5 in dizygoti (DZ) same-sex twins, while r is the orrelation between C, whih is 1 in MZ twins and DZ same-sex twins. r g and r an be estimated for DZ opposite-sex twins. Regression oeffiients of observed variables on the different latent fators are shown in lower ase: h: additive geneti effet; : ommon environmental effet; e: unique environmental effet; k: salar fator. Heritabilities are onstrained to be equal aross sexes in this model, but total varianes may be different. All (nonstandardised) variane omponents for females were onstrained to be equal to a salar multiple, k 2, of the male variane omponents. Dominane geneti influene was also tested but is omitted to simplify the diagram. Statistial software Effets of sex and other ovariates on mean values were tested by generalised estimating equations (GEE). GEEs take the nonindependene between twins into aount and yields unbiased standard errors and p-values [17]. Data handling, preliminary analyses, and GEEs were performed with STATA software (StataCorp., College Station, TX, USA). Polyhori orrelations were alulated and geneti modelling performed with Mx software (Virginia Institute for Psyhiatri and Behavioral Genetis, Virginia Commonwealth University, Rihmond, VA, USA), a omputer program speifially designed for the analysis of twin and family data [18]. RESULTS Table 1 shows the harateristis of the male and female twins. The mean age of the population was 22.5 yrs (range yrs). The total number of asthmatis was 108 individuals. As shown in table 1, age, body mass index and urrent smoking status were omparable in males and females. Males were taller and heavier than females and showed signifiantly higher FEV1, FVC and PC20 values but lower FEV1/FVC values ompared with females. Females reported more often to r e FIGURE 2. Geneti and environmental orrelations and fator loadings of the best fitting bivariate model for any two asthma-related traits. For larity, only one twin is depited. Fator loadings (or path oeffiients) are expressed as square roots to make lear that squaring of those fator loadings yields estimates of geneti and environmental variane omponents, as shown in the text. G: additive geneti fator; h 2 : heritability; e 2 : unique environmental variane omponent; E: unique environmental fator; r g : geneti orrelation, r e : environmental orrelation. have dyspnoea at night and used mediation more often than males. None of the traits showed signifiant differenes between MZ and DZ twins. Table 2 presents the twin orrelations of the objetive intermediate asthma phenotypes by zygosity groups. Twin orrelations were ollapsed aross sex, beause heritability estimates were not signifiantly different between males and females (see below and table 3). MZ orrelations were onsistently higher than DZ orrelations, indiating an important ontribution of geneti fators. Parameter estimates and 95% CIs of these best-fitting models are presented in table 3. All univariate models of the eight objetive asthma-related phenotypes showed signifiant heritabilities (h 2 range: 47 83%). A signifiant salar sex effet (fig. 1) was found for FEV1, with males showing larger variability than females. Subsequently, we performed bivariate model fitting to estimate ross-trait orrelations and investigate to what extent these orrelations an be explained by shared geneti or shared environmental fators influening both phenotypes (table 4 and fig. 3). There were negative ross-trait orrelations between lung funtion variables and all other traits. That is, atopy and BHR were related to lower lung funtion levels. Phenotypi orrelations ranged (irrespetive of the sign). The strongest orrelations were among serum total IgE, speifi IgE and SPT (range ). There were nonsignifiant orrelations of eosinophil ounts with FEV1 (-0.13) and FEV1/FVC (-0.12), and of SPT with FEV1/FVC (-0.16). 264 VOLUME 36 NUMBER 2 EUROPEAN RESPIRATORY JOURNAL

5 T. WU ET AL. ASTHMA TABLE 2 Twin orrelations by zygosity of objetive intermediate asthma phenotypes Measures MZ DZ TABLE 3 Phenotype Parameter estimates and 95% CIs of best-fitting models for univariate analyses of objetive intermediate asthma phenotypes Variane omponents Pairs n FEV FVC FEV1/FVC Total serum IgE Serum speifi IgE Eosinophils Methaholine PC SPT MZ: monozygoti; DZ: dizygoti; FEV1: fored expiratory volume in 1 s; FVC: fored vital apaity; Ig: immunoglobulin; PC20: provoative onentration ausing a 20% fall in FEV1; SPT: skin prik test. FEV1, FVC and FEV1/FVC were adjusted for age, sex, height and smoking; total IgE, eosinophils and PC20 were adjusted for age and sex. Pearson orrelations were used for ontinuous traits (FEV1, FVC, FEV1/FVC, total IgE and eosinophils) and polyhori orrelations were used for the ordinal variables (PC20, speifi IgE and SPT). Reoding of PC20: 0,5, 05160, larger values indiating higher bronhial responsiveness. Coding of SPT: 0,5, larger values indiating more positive responses to the allergens. Coding of speifi IgE: 0,3, larger values indiating more positive responses to the allergens. Serum total IgE showed signifiant geneti orrelations with all other traits (range ). Speifi IgE also presented high geneti orrelations with most traits (range ) and espeially a very high orrelation with SPT (0.98). Similarly, substantial geneti orrelations of FEV1 were observed with FEV1/FVC (0.52), PC20 (-0.51), speifi IgE (-0.46) and SPT (-0.43). By ontrast, eosinophil ounts showed signifiant geneti orrelations with total IgE (0.62) and speifi IgE (0.50) only. Compared with geneti effets, fewer environmental orrelations were signifiant. FEV1 and FEV1/FVC were environmentally orrelated (0.43), as were FEV1/FVC and total IgE (-0.31), and SPT and PC20 (0.44). Total IgE, speifi IgE and SPT showed onsiderable environmental overlap with eah other (range ). Figure 3 presents the results that deompose the phenotypi orrelation into ommon geneti and environmental fators. These results onfirm the expetation that muh of the phenotypi orrelations between these objetive asthmarelated traits are attributable to geneti fators, exept for the orrelation between PC20 and SPT, whih is mainly attributable to environmental fators. DISCUSSION The aim of this study was to estimate the relative influene of geneti and environmental fators on individual differenes in eight objetive intermediate asthma phenotypes inluding baseline lung funtion (FEV1, FVC and FEV1/FVC), BHR, positive SPT, serum total and speifi IgE, and eosinophil ounts of importane, we investigated environmental and geneti overlap between these phenotypes in a subsample of young h 2 (95%CI) e 2 (95%CI) FEV1 # 0.83 ( ) 0.17 ( ) FVC 0.72 ( ) 0.28 ( ) FEV1/FVC 0.61 ( ) 0.39 ( ) Total serum IgE 0.75 ( ) 0.25 ( ) Serum speifi IgE 0.60 ( ) 0.40 ( ) Eosinophils 0.52 ( ) 0.48 ( ) PC ( ) 0.53 ( ) SPT 0.56 ( ) 0.44 ( ) h 2 : heritability; e 2 : unique environmental variane omponent; FEV1: fored expiratory volume in 1 s; FVC: fored vital apaity; Ig: immunoglobulin; PC20: provoative onentration ausing a 20% fall in FEV1; SPT: skin prik test. FEV1, FVC and FEV1/FVC were adjusted for age, sex, height and smoking; total IgE, eosinophils and PC20 were adjusted for age and sex. PC20, SPT and speifi IgE were analysed as ordinal variables with a liability model (five thresholds for PC20 and SPT; three thresholds for speifi IgE). # : salar sex effet, male.female (k51.52). twins from the NTR. In addition to the findings of geneti effets aounting for a substantial portion (o47%) of the variation of all asthma phenotypes, the main finding of our study was that these objetive intermediate asthma phenotypes had signifiant ross-trait orrelations, whih were predominantly explained by geneti rather than by environmental fators. We observed heritabilities of 61 83% for lung funtion phenotypes; higher than in previous reports [19, 20]. This may be due to the somewhat younger age of our ohort of twins, in whih environmental fators, suh as smoke exposure may have had less influene. The urrent study is one of the few fousing on omprehensive linial markers of allergy, thereby showing that the heritability of speifi IgE is 0.60 in twins when analysing it as ordinal variable based on the sum sore of the positive responses to four allergens. Few twin studies have inluded SPT [19, 21] and heritability estimates were not onsistent (0.49 versus 0.85). We also analysed SPT as an ordinal variable and found a heritability of We did not find any sex effet on heritability of these traits, exept for a signifiant salar effet with FEV1, males showing larger variability than females. We did not find any evidene for environmental fators ontributing to intermediate phenotypes that are shared among twins of a pair, probably due to gene environment interations. It is likely that environmental risk fators trigger asthma only in subjets with a larger geneti suseptibility for allergi and asthmati diseases. Evidene for linkage between ertain hromosomes and asthma were, for example, only found in those who where exposed to igarette smoke in early hildhood; in other words, ertain genes may be expressed only upon environmental exposure [22 24]. These results emphasise the importane of onsidering environmental information in geneti studies of asthma and its related traits. EUROPEAN RESPIRATORY JOURNAL VOLUME 36 NUMBER 2 265

6 ASTHMA T. WU ET AL. TABLE 4 Geneti (below main diagonal) and environmental (above main diagonal) orrelations between objetive intermediate asthma phenotypes FEV1 FEV1/FVC Total serum IgE Serum speifi IgE Eosinophils PC20 SPT FEV FEV1/FVC Total serum IgE Serum speifi IgE Eosinophils PC SPT FEV1: fored expiratory volume in 1 s; FVC: fored vital apaity; Ig: immunoglobulin; PC20: provoative onentration ausing a 20% fall in FEV1; SPT: skin prik test. FEV1, FVC and FEV1/FVC were adjusted for age, sex, height and smoking; total IgE, eosinophils and PC20 were adjusted for age and sex. PC20, SPT and speifi IgE were analysed as ordinal variables with a liability model (five thresholds for PC20 and SPT; three thresholds for speifi IgE). Signifiant orrelations are shown in bold. A speifi fous of this paper was to explore both geneti and environmental orrelations between objetive intermediate asthma phenotypes by performing bivariate variane omponents analyses. In this way, all traits assoiated with allergy, e.g. BHR, eosinophils, total and speifi IgE, and SPT, presented signifiant ross-trait orrelations. Thus, our findings support the pathophysiologial onnetion arising from shared geneti determinants. The geneti orrelation between eosinophils and total IgE (0.62) that we found fits with the finding of geneti linkage of eosinophilia to 2q33 [25], a region previously reported to be linked with serum total IgE [26], and is also in aordane with previous findings [27] that 2q24 32 showed evidene of linkage for both eosinophils and total serum IgE in Duth families. Furthermore, the latter study showed onsiderable geneti overlap among total IgE, speifi IgE and SPT. Speifi IgE and SPT were linked to hromosome 17q25 and 22q11, and total IgE and speifi IgE to hromosome 7q11 q12 [27]. Interestingly, there were omparatively strong geneti overlaps of FEV1 with PC20 and SPT as well as FEV1/ FVC. These results are in aordane with previous findings, whih have reported that FEV1, atopy and BHR were all linked to hromosome 20q13. This linkage may result from a quantitative r p FEV1 + tf FEV1 + tige FEV1 + sige FEV1 + PC20 FEV1 + SPT FEV1 + eos tf + tige tf + sige tf + PC20 tf + eos tf + SPT tige + sige tige + eos tige + PC20 tige + SPT sige + eos sige + PC20 sige + SPT eos + PC20 eos + SPT PC20 + SPT FIGURE 3. Bivariate analyses for objetive intermediate asthma phenotypes. The height of the bars in the figure orresponds to the magnitude of the phenotypi orrelations (r p ). The bars are partitioned into omponents refleting proportions of r p that are aounted for by geneti (&) and environmental (&) fators. FEV1: fored expiratory volume in 1 s; Ig: immunoglobulin; PC20: provoative onentration ausing a 20% fall in FEV1; SPT: skin prik test; FVC: fored vital apity; tf: FEV1/FVC; tige: serum total IgE; sige: speifi IgE; eos: eosinophils. Signifiant orrelations are shown in bold. 266 VOLUME 36 NUMBER 2 EUROPEAN RESPIRATORY JOURNAL

7 T. WU ET AL. ASTHMA trait lous in this region that affets several asthma-related traits [28]. Bivariate analyses also showed that a smaller number of phenotypes had signifiant environmental orrelations. In aordane with findings of FERREIRA et al. [19], we found that there were environmental orrelations of SPT with both speifi IgE and PC20, indiating that exposure and sensitisation to aeroallergens are fundamental for development of both asthma and rhinitis. This is plausible, given the observation that inhalation of allergens to whih an individual is allergi indues a more severe hyperresponsiveness, or even renders an individual hyperresponsive during the allergi season [29 31]. However, the fat that unique geneti and environmental effets were also present for any pairs of traits (i.e. geneti or environmental orrelation were not equal to 1) indiates that they are still distint intermediate phenotypes, even though their geneti or environmental aetiology is partly the same. A strength of our study is that we have analysed nonontinuous traits (speifi IgE, PC20 and SPT) as ordinal variables (more than two ategories) using a threshold model [7, 32], whereas similar previous studies merely analysed these as binary (yes/no) traits. It is known that threshold models of ordinal traits are more powerful for deteting additive geneti or ommon environmental fators [33]. Finally, some limitations need to be onsidered. First, the ratio of MZ pairs to DZ pairs (46:57) was higher than in the general population [34]. However, oversampling of MZ twins was done on purpose to ahieve similar group sizes of MZ and DZ twins, as is ommonly applied in volunteer twin studies. Apart from a slightly lower power to detet ommon environmental fators [34], this is unlikely to have influened results or generalisability of the study. Seondly, seletive asertainment of families was present in our study. Subjets studied only ame from families with twins aged o18 yrs with at least one member (twins or their parents) reported a history of asthma, leaving us with a fration of the original sample. Although this potentially may have impat on the estimates, a previous study omparing the heritability estimates in seleted samples with those in the entire original sample [35], indiated that using the entire or seleted sample gave very similar heritabilities. In onlusion, geneti effets aount for a substantial part of the variation in all objetive intermediate asthma phenotypes. The orrelations between pairs of these traits are, to a large extent, explained by geneti effets influening both phenotypes. Environmental fators also ontributed to the linial homogeneity between asthma traits, but to a smaller extent. These findings enlarge our knowledge on the geneti and environmental origins of linial homogeneity for these objetive asthma phenotypes, and provide important lues and diretions for gene-finding studies. STATEMENT OF INTEREST None delared. REFERENCES 1 de Maro R, Cerveri I, Bugiani M, et al. An undeteted burden of asthma in Italy: the relationship between linial and epidemiologial diagnosis of asthma. Eur Respir J 1998; 11: Martinez FD, Wright AL, Taussig LM, et al. Asthma and wheezing in the first six years of life. The Group Health Medial Assoiates. N Engl J Med 1995; 332: Holgate ST. Asthma genetis: waiting to exhale. Nat Genet 1997; 15: Martinez FD. Complexities of the genetis of asthma. Am J Respir Crit Care Med 1997; 156: S117 S Los H, Koppelman GH, Postma DS. The importane of geneti influenes in asthma. Eur Respir J 1999; 14: Thomsen SF, Kyvik KO, Baker V. Etiologial relationships in atopy: a review of twin studies. Twin Res Hum Genet 2008; 11: Posthuma D, Beem AL, de Geus EJ, et al. Theory and pratie in quantitative genetis. Twin Res 2003; 6: Boomsma DI, Koopmans JR, Van Doornen LJ, et al. Geneti and soial influenes on starting to smoke: a study of Duth adolesent twins and their parents. Addition 1994; 89: Koopmans JR, van Doornen LJ, Boomsma DI. Assoiation between alohol use and smoking in adolesent and young adult twins: a bivariate geneti analysis. Alohol Clin Exp Res 1997; 21: Ferreira MAR, Zhao ZZ, Thomsen SF, et al. Assoiation and interation analyses of eight genes under asthma linkage peaks. Allergy 2009; 64: Standardization of spirometry 1987 update. Statement of the Amerian Thorai Soiety. Am Rev Respir Dis 1987; 136: Cokroft DW, Killian DN, Mellon JJ, et al. Bronhial reativity to inhaled histamine: a method and linial survey. Clin Allergy 1977; 7: Rijken B, Shouten JP, Weiss ST, et al. The distribution of bronhial responsiveness to histamine in symptomati and in asymptomati subjets. A population-based analysis of various indies of responsiveness. Am Rev Respir Dis 1989; 140: Vonk JM, Boezen HM, Postma DS, et al. Perinatal risk fators for bronhial hyperresponsiveness and atopy after a follow-up of 20 years. J Allergy Clin Immunol 2004; 114: MaGregor AJ, Snieder H, Rigby AS, et al. Charaterizing the quantitative geneti ontribution to rheumatoid arthritis using data from twins. Arthritis Rheum 2000; 43: Neale MC, Cardon LR. Methodologies for geneti studies of twins and families. Dordreht, Kluwer Aademi Publishers, Tregouet DA, Duimetiere P, Tiret L. Testing assoiation between andidate-gene markers and phenotype in related individuals, by use of estimating equations. Am J Hum Genet 1997; 61: Neale MC, Boker SM, Xie G, et al. Mx: Statistial Modeling. Department of Psyhiatry, Virginia Commonwealth University, Rihmond, Ferreira MA, O Gorman L, Le Souef P, et al. Variane omponents analyses of multiple asthma traits in a large sample of Australian families asertained through a twin proband. Allergy 2006; 61: Thomsen SF, Ferreira MA, Kyvik KO, et al. A quantitative geneti analysis of intermediate asthma phenotypes. Allergy 2009; 64: Thomsen SF, Ulrik CS, Kyvik KO, et al. Multivariate geneti analysis of atopy phenotypes in a seleted sample of twins. Clin Exp Allergy 2006; 36: van Beijsterveldt CE, Boomsma DI. Genetis of parentally reported asthma, ezema and rhinitis in 5-yr-old twins. Eur Respir J 2007; 29: Colilla S, Niolae D, Pluzhnikov A, et al. Evidene for gene environment interations in a linkage study of asthma and smoking exposure. J Allergy Clin Immunol 2003; 111: Meyers DA, Postma DS, Stine OC, et al. Genome sreen for asthma and bronhial hyperresponsiveness: interations with passive smoke exposure. J Allergy Clin Immunol 2005; 115: EUROPEAN RESPIRATORY JOURNAL VOLUME 36 NUMBER 2 267

8 ASTHMA T. WU ET AL. 25 Evans DM, Zhu G, Duffy DL, et al. Major quantitative trait lous for eosinophil ount is loated on hromosome 2q. J Allergy Clin Immunol 2004; 114: Howard TD, Postma DS, Hawkins GA, et al. Fine mapping of an IgE-ontrolling gene on hromosome 2q: Analysis of CTLA4 and CD28. J Allergy Clin Immunol 2002; 110: Koppelman GH, Stine OC, Xu J, et al. Genome-wide searh for atopy suseptibility genes in Duth families with asthma. J Allergy Clin Immunol 2002; 109: Ferreira MA, O Gorman L, Le Souef P, et al. Robust estimation of experimentwise P values applied to a genome san of multiple asthma traits identifies a new region of signifiant linkage on hromosome 20q13. Am J Hum Genet 2005; 77: Kurt E, Bavbek S, Aksu O, et al. The effet of natural pollen exposure on eosinophil apoptosis and its relationship to bronhial hyperresponsiveness in patients with seasonal allergi rhinitis. Ann Allergy Asthma Immunol 2005; 95: van der Heide S, de Monhy JG, de Vries K, et al. Seasonal variation in airway hyperresponsiveness and natural exposure to house dust mite allergens in patients with asthma. J Allergy Clin Immunol 1994; 93: Booij-Noord H, de Vries K, Sluiter HJ, et al. Late bronhial obstrutive reation to experimental inhalation of house dust extrat. Clin Allergy 1972; 2: Faloner DS. Introdution to quantitative genetis. 3rd Edn. Harlow, Longman, Neale MC, Eaves LJ, Kendler KS. The power of the lassial twin study to resolve variation in threshold traits. Behav Genet 1994; 24: Hur YM, MGue M, Iaono WG. Unequal rate of monozygoti and like-sex dizygoti twin birth: evidene from the Minnesota Twin Family Study. Behav Genet 1995; 25: Behger TM, Boomsma DI, Koning H. A limited dependent variable model for heritability estimation with non-random asertained samples. Behav Genet 2002; 32: VOLUME 36 NUMBER 2 EUROPEAN RESPIRATORY JOURNAL